I Built a JSON Fixer Because I Was Tired of Counting Characters

Emiliano Saurin — Mon, 13 Apr 2026 21:05:00 +0000

JSON.parse() gives you an error at "position 42". Great. Position 42 of a 500-line config file. Let me just count characters manually, I guess.

I hit this so often that at some point I stopped counting and built a JSON Linter that shows errors with actual line numbers. But more on that in a second. First, let's talk about why JSON keeps breaking in the same ways.

It's Always the Same Five Things

Trailing commas. You delete the last field, forget the comma on the one above it. Or you copy from JavaScript, where trailing commas are fine. JSON disagrees.

{
  "name": "Alice",
  "age": 32,
}

Single quotes. A Python habit. Or a "I'll just type this real quick" habit. JSON wants double quotes, always.

Comments. JSON has no comments. None. This is honestly a spec failure. tsconfig.json and VS Code settings use JSONC precisely because people need comments in config files. But standard JSON? Nope.

{
  // this breaks everything
  "url": "https://api.example.com"
}

Unquoted keys. Valid JavaScript, invalid JSON. { name: "Alice" } fails.

Missing brackets. Usually from truncated logs or half-pasted API responses.

These five cover maybe 90% of all JSON parse errors I've ever seen.

So I Built a Fixer

The JSON Linter and Fixer does two things:

Paste broken JSON, hit "Fix" and it auto-repairs trailing commas, single quotes, comments, unquoted keys. It uses jsonrepair under the hood. It won't magically reconstruct corrupted data, but if the JSON is almost right (and it usually is), it'll clean it up.

Hit "Lint" and you get properly formatted output with syntax highlighting. Useful even when the JSON isn't broken, just minified into an unreadable blob.

{"users":[{"name":"Alice","age":32,"addresses":[{"city":"NYC","zip":"10001"}]},{"name":"Bob","age":25,"addresses":[{"city":"LA","zip":"90001"}]}]}

Nobody can read that. One click and it's formatted.

Everything runs client-side. No backend, nothing uploaded anywhere.

The Other Half: jq Without Installing Anything

Once the JSON is valid, you usually need to pull something out of it. "Give me all the user names." "Which items have status failed?" "Group these by category."

You could write a quick script. Or you could use jq, which does this kind of thing in one line. Except now you need to install jq, and you're on a machine where you can't, or you don't feel like it.

So the same site has a jq playground. Paste JSON, write a filter, see the result. Browser-only, same as the linter.

Some examples that come up all the time, given this data:

[
  { "name": "Alice", "age": 32, "role": "admin", "active": true },
  { "name": "Bob", "age": 25, "role": "user", "active": false },
  { "name": "Carol", "age": 28, "role": "admin", "active": true }
]

Pull all names:

map(.name)

Filter active admins over 30:

.[] | select(.role == "admin" and .age > 30)

Reshape into a different format:

map({ id: .name, isAdmin: (.role == "admin") })

Group by role and count:

group_by(.role) | map({ role: .[0].role, count: length })

That last one gives you [{"role": "admin", "count": 2}, {"role": "user", "count": 1}]. The kind of thing you'd normally open Python for.

Other one-liners I use constantly:

[.[] | .category] | unique           # unique values from an array
max_by(.price)                       # most expensive item
[.users[].addresses[].city] | unique # flatten and dedupe nested arrays

You get the idea. Try it with your own data.

The Workflow

I keep the linter bookmarked. API returns garbage, paste, fix, lint. Need to dig into the response, switch to the jq tab. It handles maybe 80% of my "what's in this JSON?" moments without opening a terminal.

The whole thing is open source on GitHub if you want to look at the code or contribute.

Part of echoValue. Free, no signups, no tracking.

InstaTrack: Track Your Instagram Followers & Following (Privacy-First, Open Source)

Emiliano Saurin — Mon, 06 Oct 2025 13:29:12 +0000

After getting frustrated with Instagram analytics apps that require account access or send data to their servers or are not free I built InstaTrack: a fully client-side tool to analyze followers/following.

What it does:

Upload Instagram JSON exports (followers/following)
See who doesn't follow you back and vice versa
Track trends over time with multiple snapshots

All processing happens in your browser (localStorage), zero server involvement
Try it: https://instatrack.njoylab.com
GitHub: https://github.com/njoylab/instatrack

I have few followers, so haven't tested performance with large datasets (10k+ followers). If anyone tries it with a big following, would love feedback on how it handles.

Open to any feedback on UX or features!

[iOS] Built an offline vault app - need honest UI/UX feedback

Emiliano Saurin — Fri, 15 Aug 2025 10:07:23 +0000

Hey folks,

I recently shipped “In My Pocket”

https://apps.apple.com/us/app/in-my-pocket-offline-vault/id6444294006

Main idea:

100% offline storage, no cloud, no accounts, just local data
Store key/value pairs like IBANs, loyalty cards, access codes, etc.
Core features are free, a $3 one-time IAP unlocks a few extras

Where I need help:
I deliberately went with stock iOS components for speed and familiarity, but the result feels very plain. I’m looking for constructive UI/UX suggestions to make it cleaner and more pleasant without adding bloat.

Next update: import/export between devices

Would love if you could check it out and tell me what you’d change. Brutal honesty welcome.